How to Read a Certificate of Analysis

What HPLC purity actually measures#

High-performance liquid chromatography separates the components of a sample by pumping it through a column packed with a stationary phase, so different molecules travel through the column at different rates and reach the detector at different times. What comes out the far end is a chromatogram, a plot of detector signal against time, and each component that reaches the detector produces a peak. The area under a given peak is proportional to how much of that component passed through, and the purity figure printed on most certificates is that peak's area expressed as a percentage of the total area under every peak in the run.

The detector matters as much as the column. Most routine HPLC purity work uses ultraviolet absorbance at a fixed wavelength, commonly in the low 200 nanometre range for the peptide bond or a compound-specific wavelength where the molecule carries a stronger chromophore. A detector only registers what absorbs at the wavelength it is set to. A trace impurity that does not absorb there passes through the column and past the detector without producing a peak at all, and it is never counted in the total area, so it cannot lower the reported purity figure. The number on the certificate describes the composition of what the detector saw, not the composition of the vial.

This is the limitation worth carrying into every reading of a purity figure. A 99 percent purity result states that, of the material the detector registered, 99 percent eluted as a single peak at a given retention time. It does not, by itself, state what that peak is. Two different compounds of similar polarity can produce a clean, symmetrical peak on the same column and gradient, and a chromatogram alone offers no way to distinguish between them beyond retention time, which is itself a comparison against a reference run rather than an independent proof of identity. A purity figure is a statement about proportion among what was detected, not a statement about identity.

What mass spectrometry confirms that purity alone does not#

Mass spectrometry measures the mass-to-charge ratio of ionised molecules after they leave the chromatography column, and from that ratio an instrument calculates the molecular weight of whatever produced a given peak. Where HPLC alone answers a question about proportion, mass spectrometry answers a question about identity: does the mass associated with this peak match the mass expected for the compound named on the certificate. Run as a combined method, the two measurements are taken from the same eluting peak in the same analytical run, so the purity figure and the identity confirmation refer to the same material rather than to two separate, unlinked assertions.

The reason this matters is straightforward once stated. A sample can be genuinely, instrumentally 99 percent pure by chromatographic area and still be the wrong compound entirely. Purity is a relative measurement, the main peak weighed against every other peak in the same run, and a batch that is uniformly one substance reads as highly pure regardless of what that substance is. A synthesis error, a mislabelled starting material, or a substituted compound that happens to purify cleanly will all produce an impressive purity number while saying nothing about whether the vial contains what its label claims. Mass spectrometry ties the number to the label, because a molecular weight is a physical property of the compound itself rather than a claim that depends on what the seller wrote on the certificate.

A certificate that reports purity without any accompanying identity confirmation is a weaker document than its clean purity figure suggests, and it is worth reading as answering only half the relevant question. It tells a reader that the tested lot is chromatographically consistent with itself. It does not tell a reader that the lot is the compound named on the label. The two figures address different failure modes, and a report that only ever shows one of them has, by construction, left the identity question unexamined.

How to follow a verification link back to the laboratory#

A verification link is only useful if it leads somewhere the seller does not control. An independent laboratory report typically carries a report or sample identifier unique to the tested lot, the dates the sample was received and tested, the name of the testing facility, a description of the method used (the column, the mobile phase, the detector wavelength, or the mass spectrometer configuration), and the chromatogram and spectrum themselves rather than a summary of what they showed. A report missing several of these is not automatically fraudulent, laboratories vary in how much they publish, but a report missing all of them gives a reader nothing to check.

Where that report is hosted matters as much as what it contains. A file hosted only on the seller's own domain, reachable only through a link the seller supplies and nowhere else, puts the seller in control of both the claim and the evidence for it. Nothing about that arrangement stops the same seller from posting whatever document they choose, replacing it later, or building a page that resembles a laboratory report without one ever having been issued. A report that can instead be retrieved directly from the laboratory's own domain, through a lookup tool the lab operates independently of any single customer, separates the claim from the evidence in a way a vendor-hosted file cannot.

Once a report resolves to what looks like the laboratory's own site, a few checks are worth making on that side of the link. Search the laboratory's name independently of the seller and see whether it appears in contexts unconnected to that one vendor: other clients, industry directories, accreditation bodies, published method papers. Check whether the report identifier follows a format consistent with other reports the same lab has issued, rather than looking like a one-off document assembled for a single occasion. Confirm the domain hosting the report actually belongs to the laboratory named on it, rather than a generic subdomain, a file-hosting service, or a domain that also hosts stores for unrelated sellers. If the laboratory publishes its own report lookup by identifier, use that path directly instead of relying only on the link a seller has provided.

What a forged or recycled certificate looks like#

None of the signs below proves forgery on its own. Laboratories are inconsistent in how they format reports, and a genuine certificate can look unusual for reasons that have nothing to do with fraud. Taken together, though, they are the pattern that recurs across certificates that turn out to be fabricated or reused.

• Reused lot numbers. The same lot or batch number appears on certificates issued for two different products, or the same number resurfaces across multiple restock cycles as though the material was never actually resubmitted for testing when new stock arrived.
• Dates that do not move. The report date stays fixed while a seller's own listing claims fresh stock, or the date printed on the certificate predates the point at which that batch was first offered for sale, meaning the paperwork existed before the product it supposedly covers did.
• Missing or unreadable instrument traces. No chromatogram or spectrum image is included at all, or the image is compressed until axis labels, retention times, and peak annotations are illegible, leaving nothing a reader can actually check against the stated purity figure.
• Absent method sections. The certificate states a purity percentage with no column type, mobile phase, detector wavelength, or instrument model given anywhere. A purity figure without a stated method cannot be evaluated against anything and cannot be reproduced or challenged.
• A laboratory name with no traceable existence. Searching the laboratory's name independently turns up nothing beyond pages the seller itself controls: no separate address, no accreditation record, no other client referencing the same facility, no method papers or citations anywhere outside that one seller's site.
• Cropped images that remove identifying detail. A report photographed or scanned with the header, footer, laboratory logo, or report identifier cut out of frame, leaving only a chromatogram or a results table that cannot be traced back to any specific issuing source.
• Inconsistency between the stated compound and the reported mass. The certificate names one compound in its header, but the molecular weight printed in the mass spectrum result corresponds to a different substance, the kind of mismatch that appears when a document is assembled from a template rather than generated from an actual instrument run.

What a certificate cannot tell you#

A certificate that is genuinely issued, correctly formatted, and independently verifiable still only describes the sample the laboratory actually tested. It does not establish an unbroken chain of custody between that sample and the vial a buyer eventually receives, and nothing about a valid report prevents a seller from testing one batch and shipping material from a different one under the same paperwork. Reading a certificate answers a question about a sample. It does not, by itself, answer a question about the specific unit that arrives.

Chromatographic purity and mass confirmation are both measurements of chemical composition, not confirmation of a molecule's three-dimensional structure or how it behaves once handled. A peptide, for instance, is a chain of covalently linked residues whether folded correctly or not, and a standard purity and mass workup will not distinguish a properly folded chain from a scrambled one with the same sequence and the same mass. Compositional testing and structural verification are different questions, and a certificate that addresses the first is not answering the second.

Method scope also limits what any single figure can rule out. A purity method built around one detection wavelength, or a mass method scanning one range, will not register anything outside what it was configured to look for. Material that falls outside the assay's design will not appear on the report regardless of how carefully the stated method was run, because the method was never asked to look for it. A certificate answers the question it was designed to ask, and a reader who treats it as answering every possible question about the material has misread what analytical testing does.

None of this is an argument against reading certificates carefully. A genuine, well-documented report with a working verification path is meaningfully better evidence than an unsupported claim. It is a reminder that a certificate is bounded evidence about a sample, not a guarantee about every unit sold under a given lot number, and a sceptical reader should hold it exactly that firmly and no more.